sched/cache: Keep nr_pref_llc_running in the runnable domain, to fix LLC mis-scheduling buggrafted

alb_break_llc() decides whether to break LLC preference during active
load balance. It does so by testing that every runnable fair task on the
source rq prefers its LLC:

	env->src_rq->nr_pref_llc_running == env->src_rq->cfs.h_nr_runnable

But the two counters cover different sets. nr_pref_llc_running is updated
in account_llc_enqueue()/account_llc_dequeue(), next to cfs_rq->nr_queued,
so it follows queued tasks. h_nr_runnable is updated in set_delayed()/
clear_delayed() and drops delay-dequeued tasks.

So under DELAY_DEQUEUE, a preferring task that goes to sleep stays counted
in nr_pref_llc_running while h_nr_runnable falls. The equality then breaks,
alb_break_llc() returns false, and active balance is free to pull a task
off its preferred LLC. Active balance only moves runnable tasks, and this
is the only LLC check it consults: once the stopper runs, LBF_ACTIVE_LB
skips the per-task test in can_migrate_task(). The runnable set is the one
we want.

Fix it on the counter side. A task should be counted in
nr_pref_llc_running exactly while it is both queued on its preferred LLC
(pref_llc_queued) and runnable (!sched_delayed). Define that membership
once in task_pref_llc_runnable(), and adjust the counter only through
pref_llc_running_inc()/pref_llc_running_dec() from the four sites that
change either input: account_llc_enqueue(), account_llc_dequeue(),
set_delayed() and clear_delayed(). Gating every update on the same
predicate keeps the delay, wake and dequeue paths from double-counting
or underflowing; see the comments at those sites for the ordering.

nr_llc_running and sd->llc_counts are not touched and stay on queued
semantics.

Fixes: 714059f79ff0 ("sched/cache: Handle moving single tasks to/from their preferred LLC")
Closes: https://lore.kernel.org/lkml/20260827135000.735138-1-zhanxusheng@xiaomi.com/
Reported-by: Zhan Xusheng <zhanxusheng@xiaomi.com>
Suggested-by: Chen Yu <yu.c.chen@intel.com>
Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Kayra Cizmeci <kayracizmeci@gmail.com>
Cc: <stable@kernel.org> # v7.2.x
Link: https://patch.msgid.link/06af61afedac32e6477f57feb4d658f6c411c3af.1790035273.git.tim.c.chen@linux.intel.com

1 files changed, 136 insertions, 0 deletions

diff --git a/include/asm-generic/video.h b/include/asm-generic/video.h
new file mode 100644
index 000000000..b1da2309d
--- /dev/null
+++ b/include/asm-generic/video.h
@@ -0,0 +1,136 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#ifndef __ASM_GENERIC_VIDEO_H_
+#define __ASM_GENERIC_VIDEO_H_
+
+/*
+ * Only include this header file from your architecture's <asm/fb.h>.
+ */
+
+#include <linux/io.h>
+#include <linux/mm_types.h>
+#include <linux/pgtable.h>
+#include <linux/types.h>
+
+struct device;
+
+#ifndef pgprot_framebuffer
+#define pgprot_framebuffer pgprot_framebuffer
+static inline pgprot_t pgprot_framebuffer(pgprot_t prot,
+					  unsigned long vm_start, unsigned long vm_end,
+					  unsigned long offset)
+{
+	return pgprot_writecombine(prot);
+}
+#endif
+
+#ifndef video_is_primary_device
+#define video_is_primary_device video_is_primary_device
+static inline bool video_is_primary_device(struct device *dev)
+{
+	return false;
+}
+#endif
+
+/*
+ * I/O helpers for the framebuffer. Prefer these functions over their
+ * regular counterparts. The regular I/O functions provide in-order
+ * access and swap bytes to/from little-endian ordering. Neither is
+ * required for framebuffers. Instead, the helpers read and write
+ * raw framebuffer data. Independent operations can be reordered for
+ * improved performance.
+ */
+
+#ifndef fb_readb
+static inline u8 fb_readb(const volatile void __iomem *addr)
+{
+	return __raw_readb(addr);
+}
+#define fb_readb fb_readb
+#endif
+
+#ifndef fb_readw
+static inline u16 fb_readw(const volatile void __iomem *addr)
+{
+	return __raw_readw(addr);
+}
+#define fb_readw fb_readw
+#endif
+
+#ifndef fb_readl
+static inline u32 fb_readl(const volatile void __iomem *addr)
+{
+	return __raw_readl(addr);
+}
+#define fb_readl fb_readl
+#endif
+
+#ifndef fb_readq
+#if defined(__raw_readq)
+static inline u64 fb_readq(const volatile void __iomem *addr)
+{
+	return __raw_readq(addr);
+}
+#define fb_readq fb_readq
+#endif
+#endif
+
+#ifndef fb_writeb
+static inline void fb_writeb(u8 b, volatile void __iomem *addr)
+{
+	__raw_writeb(b, addr);
+}
+#define fb_writeb fb_writeb
+#endif
+
+#ifndef fb_writew
+static inline void fb_writew(u16 b, volatile void __iomem *addr)
+{
+	__raw_writew(b, addr);
+}
+#define fb_writew fb_writew
+#endif
+
+#ifndef fb_writel
+static inline void fb_writel(u32 b, volatile void __iomem *addr)
+{
+	__raw_writel(b, addr);
+}
+#define fb_writel fb_writel
+#endif
+
+#ifndef fb_writeq
+#if defined(__raw_writeq)
+static inline void fb_writeq(u64 b, volatile void __iomem *addr)
+{
+	__raw_writeq(b, addr);
+}
+#define fb_writeq fb_writeq
+#endif
+#endif
+
+#ifndef fb_memcpy_fromio
+static inline void fb_memcpy_fromio(void *to, const volatile void __iomem *from, size_t n)
+{
+	memcpy_fromio(to, from, n);
+}
+#define fb_memcpy_fromio fb_memcpy_fromio
+#endif
+
+#ifndef fb_memcpy_toio
+static inline void fb_memcpy_toio(volatile void __iomem *to, const void *from, size_t n)
+{
+	memcpy_toio(to, from, n);
+}
+#define fb_memcpy_toio fb_memcpy_toio
+#endif
+
+#ifndef fb_memset
+static inline void fb_memset_io(volatile void __iomem *addr, int c, size_t n)
+{
+	memset_io(addr, c, n);
+}
+#define fb_memset fb_memset_io
+#endif
+
+#endif /* __ASM_GENERIC_VIDEO_H_ */